OBJECTIVE: To review the mechanism and cause of hemostatic defects following cardiopulmonary bypass (CPB) and determine the safety and efficacy of antifibrinolytic agents for use in cardiac surgery and patients likely to benefit. DATA SOURCES: A MEDLINE search (1966 to present) of the English-language literature pertaining to aminocaproic acid (ACA), tranexamic acid (TA), and aprotinin was performed. Additional literature was obtained from reference citations of pertinent articles identified through the search. STUDY SELECTION AND DATA EXTRACTION: While all articles of relevance were considered for inclusion, this review evaluates only clinical trials with emphasis on prospective, randomized, controlled studies. DATA SYNTHESIS: In reported trials, ACA and TA each reduce mediastinal blood losses by about one-third, while transfusion needs remain unchanged. ACA and TA dosing inconsistencies, omission of transfusion criteria, and unidentified surgical risk factors prevent optimal findings. Thromboembolic complications could not be ascribed to either ACA or TA in more than 950 patients studied. Aprotinin decreased mean mediastinal blood losses by 42%, 67% and 48% in primary coronary artery bypass grafting (CABG), reoperative CABG, and in CABG patients receiving aspirin, respectively. Transfusion needs were reduced 42% in primary CABG patients and 55% to 88% in high-risk patients. Patients at high risk of bleeding (i.e., reoperative CABG and patients on aspirin) demonstrated greater transfusion needs and blood loss than primary CABG patients. As blood conservation measures may eliminate the need for transfusions among primary CABG patients, patients at higher risk may benefit most from the addition of antifibrinolytic prophylaxis. CONCLUSION: The efficacy of all antifibrinolytic agents in cardiac surgery has been established, but comparative data is inconclusive to suggest an agent of choice. Thromboembolic complications have been rare and difficult to ascribe to antifibrinolytic agents. Future trials comparing efficacy of agents in high risk patients and rigorously evaluating thromboembolic events will allow unconditional recommendations.